Use of Maternal and Neonatal Health Services Associated with

Neonatal Death in Myanmar

Further Analysis of the Myanmar Demographic and Health Survey 2015-16

DHS Further Analysis Reports No. 125

DHS Further Analysis Reports No. 125

Use of Maternal and Neonatal Health Services Associated with Neonatal Death in Myanmar

Further Analysis of the Myanmar Demographic and Health Survey 2015-16

Ko Ko Zaw1 Myo Myo Mon2

Kerry L. D. MacQuarrie3

ICF Rockville, Maryland, USA

April 2019

1 University of Public Health, Department of Epidemiology, Yangon, Myanmar 2 Department of Public Health, Maternal and Reproductive Health Division, Nay Pyi Taw, Myanmar 3 The DHS Program, Avenir Health, Rockville, Maryland, USA Corresponding author: Ko Ko Zaw, Department of Epidemiology, University of Public Health, Yangon, Myanmar, Phone: +9595134907, email: zawkzaw@gmail.com

Acknowledgments: We would like to express our thanks to USAID for funding support and ICF for the DHS Further Analysis Program, which is implemented by The Demographic and Health Surveys Program. We especially thank Dr. Thet Thet Mu, Deputy Director-General (HMIS), Department of Public Health, Ministry of Health and Sports, for facilitating the entire process from the very beginning. We appreciate the co-facilitator, Dr. Elma Laguna, for her assistance. Special thanks also to Professor Khay Mar Mya, Rector of The University of Public Health, Yangon, for the kind support of our research.

Editor: Diane Stoy Document Production: Natalie Shattuck

This report presents findings from a further analysis undertaken as part of the follow-up to the 2015-16 Myanmar Demographic and Health Survey (MDHS). ICF provided technical assistance for the project. This report is a publication of The DHS Program, which is designed to collect, analyze, and disseminate data on fertility, family planning, maternal and child health, nutrition, and HIV/AIDS. Funding was provided by the U.S. Agency for International Development (USAID) through the DHS Program (#AID-OAA-C-13-00095). The opinions expressed here are those of the authors and do not necessarily reflect the views of USAID and other cooperating agencies.

The 2015-16 MDHS was implemented by the Ministry of Health and Sports of the Republic of the Union of Myanmar. Funding for the survey was provided by USAID and the Three Millennium Development Goal Fund (3MDG). ICF provided technical assistance through The DHS Program, which assists countries in the implementation of population and health surveys in countries worldwide.

The DHS Program assists countries worldwide in the collection and use of data to monitor and evaluate population, health, and nutrition programs. Additional information about The DHS Program can be obtained from ICF, 530 Gaither Road, Suite 500, Rockville, MD 20850, USA; telephone: +1 301-407-6500; fax: +1 301-407-6501; email: info@DHSprogram.com; internet: www.DHSprogram.com.

Recommended citation:

Zaw, Ko Ko, Myo Myo Mon, and Kerry L. D. MacQuarrie. 2019. Use of Maternal and Neonatal Health Services Associated with Neonatal Death in Myanmar. Further Analysis of the Myanmar Demographic and Health Survey 2015-16. DHS Further Analysis Reports No. 125. Rockville, Maryland, USA: ICF.

iii

CONTENTS

TABLES ........................................................................................................................................................ v FIGURES .................................................................................................................................................... vii ABSTRACT.................................................................................................................................................. ix ACRONYMS AND ABBREVIATIONS ........................................................................................................ xi

1 INTRODUCTION .............................................................................................................................. 1 1.1 Background ......................................................................................................................... 1 1.2 Conceptual Framework ....................................................................................................... 2 1.3 Research Objectives ........................................................................................................... 3

2. DATA AND METHODS ................................................................................................................... 5 2.1 Data ..................................................................................................................................... 5 2.2 Key Variables and Measurements ...................................................................................... 5

2.2.1 Household characteristics ...................................................................................... 5 2.2.2 Maternal characteristics ......................................................................................... 6 2.2.3 Neonatal characteristics ......................................................................................... 6 2.2.4 Health service use characteristics ......................................................................... 6

2.3 Statistical Analysis .............................................................................................................. 6

3. RESULTS ........................................................................................................................................ 9 3.1 Distribution of Deaths by Age ............................................................................................. 9 3.2 Neonatal Deaths among the Last Children Born in the Last 5 Years ............................... 10 3.3 Background Characteristics .............................................................................................. 10 3.4 Bivariate Associations with Neonatal Death ..................................................................... 14

3.4.1 Household, maternal, and neonatal characteristics and neonatal death ............. 14 3.4.2 Health service use characteristics and neonatal death ....................................... 16

3.5 Multivariate Determinants of Neonatal Death ................................................................... 17 3.5.1 Household, maternal, and neonatal determinants ............................................... 17 3.5.2 Health service use determinants .......................................................................... 17

3.6 Multivariate Determinants of Postnatal Care .................................................................... 19

4. DISCUSSION AND CONCLUSIONS ............................................................................................ 21 4.1 Discussion ......................................................................................................................... 21 4.2 Limitations ......................................................................................................................... 22 4.3 Conclusion ........................................................................................................................ 22

REFERENCES ............................................................................................................................................ 25

APPENDIX .................................................................................................................................................. 27

v

TABLES

Table 1 Household, maternal, and neonatal characteristics of the last children under age 5 born to interviewed women ........................................................................... 11

Table 2 Service use characteristics for the last children under age 5 born to interviewed women .................................................................................................. 13

Table 3 Bivariate analysis of household, maternal, and neonatal characteristics and neonatal death ......................................................................................................... 14

Table 4 Bivariate analysis of health service use and neonatal death .................................. 16 Table 5 Odds ratios of PNC with a skilled provider from multiple logistic regression

among the last children under age 5 born to interviewed women (n=3,444) .......... 20 Appendix Table A1 Odds ratio of neonatal death from multiple logistic regression among last

children under age 5 born to interviewed women (n=3,426) ................................... 27 Appendix Table A2 Odds ratio of neonatal death by type of PNC provider from bivariate logistic

regression (n=3,561) and multiple logistic regression (n=3,324) among last children under age 5 born to interviewed women, excluding neonatal deaths in the first 24 hours of life ........................................................................................ 28

Appendix Table A3 Odds ratio of PNC from multiple logistic regression, among last children under age 5 born to interviewed women (n=3,444) ................................................ 29

vii

FIGURES

Figure 1 Conceptual framework for factors associated with neonatal death .............................. 3 Figure 2 Distribution of under-5 deaths by age groups among children born to

interviewed women in the previous 5 years (n=188) .................................................... 9 Figure 3 Distribution of neonatal deaths by day of death among children born to

interviewed women in the previous 5 years (n=104) .................................................. 10 Figure 4 Odds ratio of neonatal death for significant household, maternal, and neonatal

factors from multiple logistic regression among the last children under age 5 born to interviewed women (n=3,426) ........................................................................ 17

Figure 5 Odds ratio of neonatal death for PNC use from multiple logistic regression among the last children under age 5 born to interviewed women (n=3,426) ............. 18

Figure 6 Odds ratio of neonatal death for PNC use from multiple logistic regression among the last children under age 5 born to interviewed women (n=3,324), excluding neonatal deaths in the first 24 hours of life ................................................ 19

ix

ABSTRACT

Neonatal death is a main contributor to the death of children under age 5. Research has identified factors that influence neonatal death such as the use of health services and other behavioral and biological factors. Of these factors, access to and use of maternal and neonatal health services is one of the most manageable with health interventions. This study aims to assess the association between the use of maternal and neonatal health services and neonatal death, and to provide information that will help direct priority public health interventions to reduce neonatal deaths.

We conducted secondary analysis of the 2015-16 Myanmar Demographic and Health Survey data. We analyzed data from 3,583 last children born to women age 15 to 49 and calculated the probability of neonatal death. With bivariate analysis and multiple logistic regression, we analyzed the association of neonatal death with the use of maternal and neonatal services and other important factors (household, maternal, and neonatal characteristics).

Compared to neonates who received postnatal care from skilled providers, the risk of neonatal death among neonates who received postnatal care from unskilled providers was more than 4 times higher and neonates without any postnatal care had a 9-fold increase in risk of neonatal death. More than 60% of neonates in the study did not receive any postnatal care. When compared with neonates with average birth size, neonates with small birth size had a 4-fold increase in risk of neonatal death, while neonates with a very small birth size had a 9-fold increase in the risk of neonatal death. The odds of receiving postnatal care from a skilled provider were nearly double among mothers who had four or more antenatal visits compared to those who received fewer than four visits or no antenatal care.

The results of this study suggest that a major focus of public health interventions should be improving the use of postnatal care. This should also be linked with the promotion of proper antenatal care and institutional delivery because these services are closely linked and integrated.

KEY WORDS: neonatal death, antenatal care, Myanmar, DHS, postnatal care

xi

ACRONYMS AND ABBREVIATIONS

ANC antenatal care

CI confidence interval

DHS Demographic and Health Survey

MDHS Myanmar Demographic and Health Survey MPHC 2014 Myanmar Population and Housing Census 2014

OR odds ratio

PNC postnatal care PSU primary sampling unit

1

1 INTRODUCTION

1.1 Background

Globally, 2.6 million children died within their first 28 days of life, the neonatal period, in 2016 with neonatal deaths accounting for nearly half (46%) of all under-5 deaths (Hug et al. 2017). The first few days of newborn life are critical because more than one-third of all neonatal deaths occur within the first 24 hours of life (Hug et al. 2017). Although there has been substantial progress in reducing the child mortality rate worldwide, the progress with neonatal mortality is slower. The mortality of children age 1-59 months declined by 62% from 1990 to 2016, while neonatal mortality declined only by 49% (Hug et al. 2017).

Worldwide, almost all neonatal deaths occurred in low and middle-income countries including Myanmar (Lawn et al. 2005). In Myanmar, according to the Demographic Health Survey (DHS), the neonatal mortality rate was 25 per 1,000 live births in the 5 years before the survey (MOHS and ICF 2017). Newborn death was major contributor to under-5 deaths, with the proportion of neonatal deaths in under-5 deaths increasing from 24% in 2003 to 48% in 2013 (MOH 2014). The most common causes of neonatal deaths in Myanmar were prematurity-related conditions or low birth weights (36%), birth asphyxia (26%), neonatal jaundice (15%), and neonatal sepsis (12%) (MOH 2014).

Health outcomes of newborns are influenced by maternal, socioeconomic, children’s, pregnancy, and childbirth characteristics, as well as health care services for the mother and newborn (Fort et al. 2008; Lawn et al. 2005). Maternal characteristics such as young maternal age, low education, short birth interval, birth order (first birth is associated with a higher mortality rate compared to second and third births), and poor maternal nutritional status increases the risk of neonatal death (MOHS and ICF 2017; Fort et al. 2008). Poverty, which is associated with poor sanitation, housing conditions, and nutrition, is an underlying cause of neonatal mortality, which can increase risk of infection and the inability to access health services (Lawn et al. 2005). Characteristics that determine newborn survival include low birth weight, the main cause of neonatal deaths (60-80%), which can be associated with pre-term birth or intrauterine growth retardation (Lawn et al. 2005).

Characteristics of pregnancy and childbirth are also important determinants of neonatal health. Maternal health conditions during pregnancy such as ante-partum hemorrhage and pregnancy-induced hypertension can affect newborn survival. Complications of childbirth such as mal-presentation of the fetus and obstructed labor can cause birth asphyxia in neonates, one of the most common causes of newborn deaths.

Health interventions that improve newborn health include proper antenatal care (ANC), delivery care, and postnatal care (PNC). Limited access to and poor use of these health services are associated with increased newborn deaths. Research has shown that effective interventions in quality ANC such as the use of insecticides-treated bed nets in pregnancy; antenatal syphilis screening combined with penicillin treatment; detection and proper treatment of pregnancy complications such as pre-eclampsia, eclampsia, and gestational diabetes; iron and nutrient supplementation for maternal under-nutrition; and antenatal steroids can reduce newborn deaths (Bhutta et al. 2014). The day of childbirth is one of the crucial times for both the mother’s and newborn’s survival. Research has found that availability of basic and comprehensive emergency obstetric and newborn care at the time of childbirth can reduce newborn deaths by 40% (Bhutta

2

et al. 2014). Immediate newborn care with clean cord care, early initiation of breastfeeding, and Kangaroo mother care are effective postnatal interventions that reduce neonatal deaths. With small and ill newborns, the availability of immediate care by skilled health personnel is also essential for improving newborn survival (Bhutta et al. 2014). Thus, skilled attendance and institutional delivery are important factors for preventing newborn deaths.

To reduce neonatal deaths, coverage of effective health interventions must be increased to reach the poorest and most underserved populations. Increasing coverage of essential maternal, newborn, and child health interventions is a priority in the Myanmar National Health Plan 2017-2021 (MOHS 2016). According to the Myanmar DHS (MDHS) 2015-16, 81% of women of reproductive age received ANC from skilled providers, while only 59% completed four antenatal visits. Only 37% of births took place at health facilities; 60% were delivered by skilled providers; and only 36% of newborns received the recommended postnatal check-up within the first 2 days after birth (MOHS and ICF 2017).

All factors associated with neonatal death must be identified in order to address the causes of preventable maternal and newborn deaths in Myanmar. Among them, maternal and neonatal health care are important determinants of neonatal survival that can improve access and use of health interventions. However, there are limited studies with national data on the determinants of neonatal deaths, and especially on the role of health services in neonatal survival in Myanmar.

The MDHS was conducted for the first time in Myanmar in 2015-16. The MDHS documented household characteristics, health conditions, and use of health services by women and men of reproductive age. Information about neonatal death and factors that influence neonatal death such as wealth quintile, maternal education, age, birth interval, and birth size were reported. Nevertheless, there are more data that affect neonatal health in this nationally representative survey such as water and sanitation conditions, maternal health conditions, and use of health services during pregnancy, delivery, and the postnatal period. This study aims to identify the maternal and neonatal health services factors associated with neonatal death in Myanmar using data from 2015-16 MDHS. The findings from this study will provide information that will help prioritize high-impact health interventions that can reduce neonatal deaths in Myanmar.

1.2 Conceptual Framework

The conceptual framework is based on a literature review of factors that influence neonatal death and data available from the MDHS. Factors were divided into four categories, namely household; maternal; neonatal; and health service factors such as ANC, delivery care, and PNC. A total of 20 possible variables include four variables related to household characteristics (urban-rural residence, wealth quintile, source of drinking water, and toilet type); four variables for maternal characteristics (age, education, smoking status, and smoker in home); five variables for neonatal characteristics (sex of the child, birth weight, previous birth interval, birth order, and type of birth—either single or multiple); three variables for maternal health services (antenatal visit, ANC attendant, and receipt of essential ANC components); three variables for delivery service (birth place, birth attendants, delivery by cesarean section); and one for the provider of PNC (if PNC was provided, and if so, by either a skilled or unskilled provider).

3

Figure 1 Conceptual framework for factors associated with neonatal death

1.3 Research Objectives

The main objective of this study is to assess the association between the use of maternal and neonatal health care and neonatal death. Specifically, this study aims to identify the association of household, maternal, and neonatal characteristics, use of maternal neonatal health services, and neonatal death.

Household characteristics

ResidenceWealth quintileSource of drinking

waterType of toilet

Postnatal characteristics

Type of provider of PNC visit

Maternal characteristics

AgeEducationActive smokingPassive smoking

Newborn characteristics

SexBirth sizePrevious birth

intervalBirth order

Neonatal Death

Antenatal characteristics

ANC visitANC attendantEssential

components of ANC

Delivery characteristics

Place of deliveryBirth attendantCesarean

sectionMultiple

pregnancy

5

2. DATA AND METHODS

2.1 Data

This study was a secondary data analysis of the 2015-16 MDHS, the first DHS conducted in Myanmar. The survey used nationally representative samples to estimate the core demographic and health indicators of the entire country. The indicators were disaggregated by seven states, eight regions, and urban versus rural areas. The total sample included 13,260 households. From these households, 16,800 women age 15-49 were interviewed for the women’s survey and 7,500 men age 15-49 for the men’s survey.

The 2015-16 MDHS used the sampling frame of the Myanmar Population and Housing Census 2014 (MPHC 2014), which included 76,990 primary sampling units (PSUs) for the entire country. Fifteen states/regions were stratified into urban and rural areas. Thirty sampling strata were created with samples selected independently from each sampling stratum. Two-stage stratified cluster sampling was used to select 442 clusters—123 from urban and 319 from rural areas. A fixed number of 30 households per cluster was selected to achieve 13,260 households. The survey did not allow replacement for any non-responsive households. The household response rate was 98% (12,500/13,238). The women’s individual response rate was 96% (12,885/13,454).

All MDHS data are publicly available free of charge in the form of standard recode data files from The DHS Program at https://www.dhsprogram.com/Data/. Since this secondary analysis aimed to identify the factors associated with neonatal death, including the use of maternal and neonatal services, we used the children’s recode data file (KR file) in the MDHS datasets because data on the use of maternal and neonatal services were available only in this data set. The children’s data file (KR file) includes data on 4,815 children under age 5 (unweighted number) of interviewed women who were age 15-49 at the time of survey. Because data on use of maternal and neonatal services were only collected for the last children born to the interviewed women, our analysis used data only on the last children, which included 3,867 in unweighted numbers and 3,583 in weighted numbers.

2.2 Key Variables and Measurements

The dependent variable is neonatal death, which is defined as death of a newborn within 28 days after birth. Because we confined our analysis to the last children born to the interviewed women under age 5, neonatal death in our study is neonatal death among the last children. The 20 independent variables in our study included household, maternal, neonatal, and service use characteristics.

2.2.1 Household characteristics

Residence is categorized as either urban or rural based on the a priori definition of the PSU in the sample frame. Household wealth quintile is based on an index of the number and types of assets and housing characteristics of the household, and then divided into quintiles of the de jure population. Source of drinking water for the household is a dichotomous variable categorized as improved or non-improved. Improved sources include piped water, public taps, standpipes, tube wells, boreholes, protected dug wells and springs, rainwater, and bottled water. Non-improved sources are all other sources, including unprotected dug well, unprotected spring, tank truck, and surface water. Type of toilet is a categorical

6

variable with shared, unimproved, and improved (not shared) options. Improved toilets include flush to piped sewer system, septic tank, or latrine; ventilated improved pit latrine; pit latrine with slab; or composting toilet. All other types are considered unimproved.

2.2.2 Maternal characteristics

Maternal age is based on the mother’s age at the time of birth and categorized as < age 20, and ages 20-29, 30-39, and 40-49. Mother’s education is grouped in no education, primary education, secondary education, and higher education. Smoking status of mother and smoker in home are dichotomous variables based on the presence or absence of smoking.

2.2.3 Neonatal characteristics

Neonatal characteristics include sex of the child (male or female). Birth size is based on the mother’s report and is categorized as very large, larger than average, average, smaller than average, and very small. Previous birth interval is a categorical variable that measures the duration of the interval preceding the birth, and is grouped into less than 2 years, 2 years, 3 years, 4 or more years, and two additional categories. The “first child” category refers to first born children for whom there is no preceding birth interval and “don’t remember” categorizes mothers who are uncertain of the duration of the preceding interval. Birth order is grouped into first order births, 2-3, 4-6, and 7 or greater.

2.2.4 Health service use characteristics

ANC visits is a dichotomous variable grouped into < 4 visits (including no ANC visits) and 4+ visits. ANC attendants are categorized as skilled (doctor, nurse, midwife, or lady health visitor) or not skilled (auxiliary midwife, community/village health worker, traditional birth attendant, or other). Receipt of essential ANC components is categorized as no ANC, ANC but not all three components, and ANC with all three components, which include measuring blood pressure, giving information about signs of pregnancy, and taking a urine or blood sample during an ANC visit. Place of delivery is grouped into either institutional (public, private, or NGO sector health facility) or non-institutional (home or other). Birth attendants were categorized as skilled or unskilled in the same manner as the ANC attendant. Delivery by cesarean section is a dichotomous variable, with type of birth grouped as either a single or multiple birth (twins or triplets). Postnatal care (PNC) is a categorical variable grouped into no PNC, PNC with a skilled provider, and PNC with an unskilled provider by using the same definition of skilled and unskilled.

2.3 Statistical Analysis

Background characteristics of the last children under age 5 born to the interviewed women were described. We describe the distribution of under-5 deaths by age to assess the contribution of neonatal deaths to all child mortality and to examine the timing of death among neonatal deaths. We conducted bivariate analysis to assess the relationship between the 20 independent variables and neonatal death. In all bivariate analyses, the independent variables were treated as categorical variables. The association between each independent variable and neonatal death was assessed by calculating the odds ratio (OR) and 95% confidence interval (CI). P values were also calculated by using Pearson χ2 test to determine the statistical significance of each association. The significance level was set at 0.05.

7

We conducted multivariate logistic regression to assess the independent association of maternal and neonatal service utilization and neonatal death (binary outcome), controlling for other independent variables simultaneously. Three variables related to the use of maternal and neonatal services were included in this regression modelling as the main independent variables: frequency of ANC visit (less than four visits or four or more visits), type of delivery (institutional or non-institutional), and PNC (no PNC, PNC by non-skilled provider, or PNC by skilled provider). Other independent variables included in this regression modelling were urban-rural residence, wealth quintile, source of drinking water, type of toilet, maternal age, maternal education, maternal smoking status and smoker in home, sex of the last child, birth size of the last child, previous birth interval of the last child, and birth order of the last child.

Our analysis of neonatal death assumes that use of health services precede our outcome—neonatal death. However, with PNC, it is possible that this assumption is violated. It could be that the lack of PNC among some neonatal deaths is due to these deaths occurring immediately after birth and before there is an opportunity to provide PNC, rather than no PNC leading to neonatal death. This bias could contribute to an overestimate of the strength of any relationship detected between PNC and neonatal death. Therefore, we conducted a sensitivity analysis to check the robustness of our findings about PNC. We repeated the regression of neonatal death with a restricted sample that omitted deaths that occurred in the first 24 hours of life. We compared the OR and p-value for the PNC variables between the full sample model and the restricted sample model.

We further analyzed determinants of PNC with a skilled provider using a multivariate logistic regression model because of this factor’s potential importance to reducing neonatal deaths. In this model, the outcome is expressed as PNC with a skilled provider versus PNC with an unskilled provider or no PNC (binary outcome). The service use, maternal, and neonatal variables specified in this model are similar to those for the model of neonatal deaths: frequency of ANC visit (less than four visits or four or more visits), type of delivery (institutional or non-institutional), urban-rural residence, wealth quintile, source of drinking water, type of toilet, maternal age, maternal education, maternal smoking status and smoker in home, sex of child, birth size, previous birth interval, and birth order.

All analyses were performed with Stata software Version 15 (Stata Corp. 2017). The complex survey design (stratification, clustering, and unequal sampling probabilities) was accounted for in every analysis by using ‘svyset’ and ‘svy: tab’ and ‘svy:logit’ Stata commands.

9

3. RESULTS

3.1 Distribution of Deaths by Age

Figure 2 shows the distribution of childhood death by age, categorized as neonatal deaths (0-28 days), post-neonatal deaths (1-11 months), and child deaths (1-5 years). The distribution indicates that neonatal deaths comprise the majority (55%) of deaths to children under age 5. This is followed by post-neonatal deaths, which are more than a third of all under age 5 deaths, and child deaths at 10%.

Figure 2 Distribution of under 5 deaths by age groups among children born to interviewed women in the previous 5 years (n=188)

Figure 3 also shows the distribution of deaths by age, specifically within the neonatal period. The majority of neonatal deaths occur in the first several days of life. Deaths within the first 24 hours of life contribute disproportionately to all neonatal deaths, constituting 39 of 104 neonatal deaths. The number of deaths declines rapidly thereafter and levels off after the third day of life. Deaths in the 0-2 days after birth account for 70 of the 104 neonatal deaths.

55%104

35%66

10%18

Neonatal(0-28 days)Post-neonatal(1-11 months)Child(1-4 years)

10

Figure 3 Distribution of neonatal deaths by day of death among children born to interviewed women in the previous 5 years (n=104)

3.2 Neonatal Deaths among the Last Children Born in the Last 5 Years

Among the 3,583 last children born to the women age 15 to 49 in the 5 years before the survey, 52 children died in the 28 days after birth. The probability of neonatal death was 1.44% (95% CI=1.03 to 2.01).

3.3 Background Characteristics

The distribution of household characteristics, maternal characteristics of the last children under 5 years born to the interviewed women, characteristics of the last children under 5 years born to the interviewed women, and service utilization characteristics for the last children under 5 years born to the interviewed women was compared between the children alive and the children dead in the 28 days after birth, as shown in Tables 1 and 2.

The distribution of household characteristics, maternal characteristics of the last children under 5 years born to the interviewed women, characteristics of the last children under 5 years born to the interviewed women, and service utilization characteristics for the last children under 5 years born to the interviewed women is compared between the children alive and the children dead during 28 days after birth in Tables 1 to 4.

A greater proportion of children who died within 28 days after birth lived in a rural residence, were in poorest and poorer wealth quintiles, and lived in households with non-improved sources of drinking water as well as shared and unimproved toilets (Table 1).

A greater proportion of children who died within 28 days after birth had mothers with no education, although the distribution of the mother’s age, mother’s smoking status, and presence of smokers at home was similar between the children alive and the children dead during the 28 days after birth (Table 1). A

11

greater proportion of children who died during the 28 days after birth were female, with a birth size larger or smaller than normal, and a birth order higher than three (Table 1).

Table 1 Household, maternal, and neonatal characteristics of the last children under age 5 born to interviewed women

Children alive during 28 days after birth

Children dead during 28 days after birth Total

Characteristic % n=3,531 % n=52 % n=3,583

Household characteristics Residence

Urban 23.5 830 15.5 8 23.4 838 Rural 76.5 2,701 84.5 44 76.6 2,744

Wealth quintile Poorest 27.3 965 32.5 17 27.4 981 Poorer 21.8 770 33.8 17 22.0 787 Middle 17.5 618 12.5 6 17.4 624 Richer 17.9 631 13.2 7 17.8 638 Richest 15.5 548 7.9 4 15.4 552

Source of drinking water Non-improved 19.6 690 27.7 14 19.7 705 Improved 80.4 2,841 72.3 37 80.3 2,878

Type of toilet Shared 9.8 347 13.0 7 9.9 354 Unimproved 47.6 1,680 55.8 29 47.7 1,708 Improved 42.6 1,504 31.1 16 42.4 1,520

Maternal characteristics Mother’s age <20 years 7.0 248 0.3 1 6.9 249 20-29 years 49.9 1,762 52.6 27 50.0 1,790 30-39 years 35.2 1,243 36.8 19 35.2 1,262 40+ years 7.8 277 10.2 5 7.9 282

Mother’s education No education 16.2 570 31.3 16 16.4 587 Primary 45.6 1,611 34.5 18 45.5 1,629 Secondary 29.9 1,056 25.7 13 29.8 1,069 Higher 8.3 294 8.6 4 8.3 298

Smoking status of mother Not current smoker 95.8 3,384 93.5 48 95.8 3,432 Current smoker 4.2 147 6.5 3 4.2 150

Smoking in home Smoker 58.1 2,052 60.2 31 58.1 2,083 No smoker 41.9 1,479 39.8 21 41.9 1,499

Continued…

12

Table 1 Continued

Children alive during 28 days after birth

Children dead during 28 days after birth Total

Characteristic % n=3,531 % n=52 % n=3,583

Neonatal characteristics Sex of child

Male 51.5 1,818 46.2 24 51.4 1,842 Female 48.5 1,713 53.8 28 48.6 1,741

Birth size Very large 2.2 74 3.9 2 2.2 76 Larger than average 23.2 788 21.7 10 23.2 799 Average 61.8 2,099 37.4 18 61.5 2,118 Smaller than average 11.4 387 28.0 14 11.6 400 Very small 1.4 48 9.0 4 1.5 53

Previous birth interval <2 years 7.4 261 9.6 5 7.4 266 2 years 10.5 372 19.5 10 10.7 382 3 years 10.9 386 14.2 7 11.0 393 4+ years 36.5 1,287 36.7 19 36.5 1,306 First child 34.4 1,214 16.6 9 34.1 1,222 Don’t remember 0.3 11 3.3 2 0.4 13

Birth order 1 34.4 1,214 16.6 9 34.1 1,222 2-3 42.9 1,513 45.8 24 42.9 1,537 4-6 18.3 644 28.8 15 18.4 659 7+ 4.5 160 8.7 4 4.6 165

13

Mothers of the children who died during the 28 days after birth were more likely to have no ANC, receive ANC with less than the three essential components of ANC, have the child delivered non-institutionally, have the child delivered by unskilled birth attendants, and received no PNC (Table 2).

Table 2 Service use characteristics for the last children under age 5 born to interviewed women

Children alive during 28 days after birth

Children dead during 28 days after birth Total

Characteristic % n=3,531 % n=52 % n=3,583

Antenatal characteristics ANC visit

<4 visit 40.7 1,438 45.0 23 40.8 1,462 4+ visit 58.6 2,070 55.0 28 58.6 2,099 Don’t know 0.6 21 0.0 0 0.6 21 Missing 0.0 1 0.0 0 0.0 1

ANC attendants Not skilled 19.2 678 26.4 14 19.3 692 Skilled 80.8 2,853 73.6 38 80.7 2,891

Essential components of ANC No ANC 12.8 451 16.7 9 12.8 460 ANC but not all 3

components 42.4 1,497 47.7 25 42.5 1,521

ANC with all 3 components 44.8 1,583 35.6 18 44.7 1,601

Delivery Characteristics Place of delivery

Institutional 40.8 1,439 28.0 14 40.6 1,453 Non-institutional 59.2 2,092 72.0 37 59.4 2,129

Birth attendant Not skilled 36.3 1,281 52.2 27 36.5 1,308 Skilled 63.7 2,250 47.8 25 63.5 2,275

Delivery by cesarean section No 80.5 2,831 91.4 47 80.6 2,878 Yes 19.5 687 8.6 4 19.4 692

Type of birth Single 99.1 3,499 92.7 48 99.0 3,547 Multiple (twins, triplets) 0.9 32 7.3 4 1.0 36

Postnatal characteristics Type of provider for first

PNC visit No PNC 51.8 1,828 81.0 42 52.2 1,870 PNC with skilled provider 13.3 471 12.9 7 13.3 477 PNC with unskilled provider 34.9 1,232 6.1 3 54.5 1,235

14

3.4 Bivariate Associations with Neonatal Death

3.4.1 Household, maternal, and neonatal characteristics and neonatal death

Crude or unadjusted association of neonatal death with household characteristics, maternal characteristics of the last children under 5 years born to the interviewed women, characteristics of the last children under 5 years born to the interviewed women, and service utilization characteristics for the last children under 5 years born to the interviewed women are presented in Tables 3 to 5.

There were no significant associations between neonatal death and any of the household characteristics we measured in this study (Table 3). Children of the mothers with no education, and who were current smokers, female children, and children with birth size larger than average all appeared more likely to die during 28 days after birth, although these associations were not statistically significant at the 0.05 level (Table 3). However, statistical significance was found in associations with birth size that was smaller or very small, and with birth order of four or higher (Table 3).

Table 3 Bivariate analysis of household, maternal, and neonatal characteristics and neonatal death

Characteristics Odds ratio 95% CI P value

Household characteristics Residence

Urban 1 Rural 1.67 [0.72, 3.86] 0.227

Wealth quintile Poorest 2.34 [0.65,8.39] 0.191 Poorer 3.05 [0.88,10.52] 0.078 Middle 1.40 [0.37,5.37] 0.622 Richer 1.46 [0.36, 5.96] 0.600 Richest 1

Source of drinking water Non-improved 1.57 [0.78,3.18] 0.206 Improved 1

Type of toilet Shared 1.81 [0.60,5.47] 0.292 Unimproved 1.61 [0.78,3.31] 0.199 Improved 1

Maternal characteristics Mother’s age

<20 years 0.04 [0.01,0.19] <0.001 20-29 years 1 30-39 years 0.99 [0.49,2.01] 0.981 40+ years 1.24 [0.45,3.39] 0.675

Mother’s education No education 1.88 [0.54,6.55] 0.323 Primary 0.73 [0.21,2.52] 0.621 Secondary 0.83 [0.24,2.90] 0.773 Higher 1

Continued…

15

Table 3 Continued

Characteristics Odds ratio 95% CI P value

Smoking status of mother Not current smoker 1 Current smoker 1.59 [0.48,5.27] 0.446

Smoking in home No smoker 1 Smoker 1.09 [0.59,2.01] 0.782

Neonatal characteristics Sex of child

Male 1 Female 1.24 [0.63,2.41] 0.535

Birth size Very large 2.97 [0.39,22.61] 0.292 Larger than average 1.54 [0.66,3.58] 0.312 Average 1 Smaller than average 4 [1.62,10.16] 0.003 Very small 10.45 [2.42,45.09] 0.002

Previous birth interval1 <2 years 1.29 [0.43,3.86] 0.648 2 years 2 [0.75,4.48] 0.180 3 years 1.29 [0.38,4.40] 0.685 4+ years 1 First child 0.48 [0.19,1.24] 0.129

Birth order 1 1 2-3 2.21 [0.86,5.64] 0.097 4-6 3.26 [1.21,8.82] 0.020 7+ 4 [1.05,15.02] 0.042

1 “Don’t remember” responses are excluded.

16

3.4.2 Health service use characteristics and neonatal death

Children of mothers with no ANC, mothers who received ANC with less than the three essential components, mothers who had the child delivered non-institutionally, mothers who had the child delivered by non-skilled birth attendants, mothers who received no PNC or PNC by unskilled providers, mothers who had the child delivered by cesarean section, and mothers who delivered multiple births were more likely to be dead during the 28 days after birth (Table 4). Statistical significance was found in associations with multiple births and with no PNC or PNC by unskilled providers (Table 4).

Table 4 Bivariate analysis of health service use and neonatal death

Characteristic Odds ratio 95% CI P value

ANC visit1 <4 visit 1.18 [0.62.2.23] 0.615 4+ visit 1

ANC attendants Not skilled 1.51 [0.74,3.07] 0.258 Skilled 1

Receipt of essential component of ANC No ANC 1.64 [0.64,4.2] 0.299 ANC but not all 3 components 1.41 [0.68,2.94] 0.35 ANC with all 3 components 1

Place of delivery Institutional 1 Non-institutional 1.77 [0.88,3.56] 0.111

Birth attendant at delivery Not skilled 1.92 [1.00,3.69] 0.05 Skilled 1

Delivery by cesarean section No 2.58 [0.83,8.01] 0.100 Yes 1

Type of birth Single 1 Multiple (twins, triplets) 8.64 [2.41,30.97] 0.001

Type of provider for first PNC visit No PNC 8.95 [2.52,31.84] 0.001 Unskilled 5.55 [1.12,27.60] 0.036 Skilled 1

1 “Don’t know” and “Missing value” responses were excluded.

17

3.5 Multivariate Determinants of Neonatal Death

3.5.1 Household, maternal, and neonatal determinants

The independent association of neonatal death with each of the household, maternal, children’s, and service use characteristics for the children was obtained by multiple logistic regression. As shown in Figure 4, the results were statistically significant. Complete results from multiple logistic regression are shown in Appendix Table A1.

Figure 4 Odds ratio of neonatal death for significant household, maternal, and neonatal factors from multiple logistic regression1 among the last children under age 5 born to interviewed women (n=3,426)

1 Only significantly associated factors for neonatal death are shown. Full model is adjusted for place of residence, wealth index, source of water, type of toilet, mother’s age at birth of the child, educational level, current smoking status of mother, smokers in the home, sex of child, birth size, previous birth interval, birth order, multiple births, number of ANC visits, place of delivery, and type of provider for the PNC visit. 3.5.2 Health service use determinants

Compared to neonates who received PNC from skilled providers, neonates without PNC had a 9-fold increase in the risk of neonatal death, while neonates with PNC from unskilled providers had a nearly 5-fold increase in the risk of neonatal death (Figure 4).

Seventy-one percent of mothers and 36% of neonates received the recommended PNC check-up within the first 2 days after birth. This means that more than 60% of neonates in the study did not receive PNC within the first 2 days after birth (not shown in the table). Compared with neonates with average birth size, neonates with small birth size had a 4-fold increase in the risk of neonatal death and neonates with a very small birth size had a 9-fold increase in the risk of neonatal death (Figure 4).

18

To show the protective effect of PNC by an unskilled or skilled provider, the multiple logistic regression was re-run with no PNC as the reference group. The result is shown below in Figure 5.

Figure 5 Odds ratio of neonatal death for PNC use from multiple logistic regression1 among the last children under age 5 born to interviewed women (n=3,426)

1 Adjusted for place of residence, wealth index, source of water, type of toilet, mother’s age at birth of the child, educational level, current smoking status of mother, smokers in the home, the sex of child, birth size, previous birth interval, birth order, multiple births, number of ANC visits, place of delivery, and the type of provider for the PNC visit. The results showed that PNC, whether provided by an unskilled or a skilled provider, is protective against neonatal death. The PNC by an unskilled provider was associated with a 50% decrease in odds of neonatal death (OR=0.54, 95% CI=0.21 to 1.35), while PNC by a skilled provider was associated with a 90% decrease in the odds of neonatal death (OR=0.11, 95% CI= 0.03 to 0.38) (Figure 5).

As a robustness check, we repeated this analysis excluding neonatal deaths in the first 24 hours of life. We assumed that PNC precedes the outcome, neonatal death versus survival, in our first analysis. However, for those deaths in the first 24 hours, it is possible that the reason for no PNC is that death occurred before the opportunity to provide PNC, and our assumed temporal order is reversed. Excluding these early neonatal deaths allows us to account for this possible confounding. Figure 6 shows that our results held, even when we excluded deaths in the first 24 hours. Compared to no PNC, the odds of neonatal death are 80% lower among infants who received PNC by a skilled provider. Further details are shown in Appendix Table A2.

19

Figure 6 Odds ratio of neonatal death for PNC use from multiple logistic regression1 among the last children under age 5 born to interviewed women (n=3,324), excluding neonatal deaths in the first 24 hours of life

1 Adjusted for place of residence, wealth index, source of water, type of toilet, mother’s age at birth of the child, educational level, current smoking status of mother, smokers in the home, the sex of child, birth size, previous birth interval, birth order, multiple births, number of ANC visits, place of delivery, and the type of provider for the PNC visit.

20

3.6 Multivariate Determinants of Postnatal Care

We further explored determinants of PNC with a skilled provider because this type of PNC emerged as an important protective factor against neonatal death in the previous analysis. Table 5 shows the odds ratios for household, maternal, neonatal, and other service use characteristics that are significantly associated with PNC with a skilled provider. Complete results from this multiple logistic regression are shown in Appendix Table A3.

Table 5 Odds ratios of PNC with a skilled provider from multiple logistic regression among the last children under age 5 born to interviewed women (n=3,444)

Characteristics Odds ratio1 95% CI

Residence Urban Ref. Rural 1.7*** [1.3, 2.3]

Mother’s education No education Ref. Primary 1.4* [1.0,1.8] Secondary 1.3 [1.0,1.8] Higher 1.6* [1.1,2.4]

Previous birth interval2 <2 years Ref. 2 years 1.3 [0.9,1.9] 3 years 1.2 [0.8,1.7] 4+ years 1.2 [0.8,1.6] The first child 1.4* [1.0,2.0]

Number of ANC visits3 <4 visits Ref. 4+ visits 1.7*** [1.4,2.4]

* p<0.05, ** p<0.01, *** p<0.001 1 Only significantly associated factors for neonatal death are shown. Full model is adjusted for place of residence, wealth index, source of water, type of toilet, mother’s age at birth of the child, educational level, current smoking status of mother, smokers in the home, the sex of child, birth size, previous birth interval, birth order, multiple births, number of ANC visits, place of delivery, type of provider for PNC visit 2 “Don’t know” responses were excluded. 3 “Don’t know” and “Missing value” responses were excluded.

Among household characteristics, only residence shows an association with PNC. Births to women living in rural areas have a 70% greater odds of receiving PNC from a skilled provider. Births for mothers with primary or higher than secondary education have 40%-60% greater odds of PNC with a skilled provider than mothers with no education. Births with no previous birth interval because they were the first born also had 40% greater odds of PNC with a skilled provider than births with a preceding birth interval of less than 2 years. No other maternal or neonatal characteristics were associated with PNC.

With use of other services, mothers who received four or more ANC visits during their pregnancy had 70% higher odds of receiving PNC from a skilled provider. No such association was found between the place of delivery and PNC from a skilled provider.

21

4. DISCUSSION AND CONCLUSIONS

4.1 Discussion

Using the 2015-16 MDHS data, we analyzed data on the 3,583 last children born to women age 15 to 49. Among these children, 52 children died in the first 28 days of their life. The overall probability of neonatal death was 1.4% (95% CI=1.0 to 2.0). The age pattern of deaths in the neonatal period and the large contribution of neonatal deaths to all under-5 deaths in this study is similar to the age pattern found elsewhere (Hug et al. 2017).

We found a strong association between the use of PNC and neonatal survival. Compared to neonates who received PNC from skilled providers, neonates without PNC had a 9-fold increase in the risk of neonatal death. More than 60% of neonates in the study did not receive PNC within the first 2 days after birth. For neonates who received PNC by an unskilled provider, the risk of neonatal death was over 4-times higher than those who received PNC from skilled providers. Inversely, PNC by unskilled providers was associated with a 50% decrease in the odds of neonatal death, while PNC by a skilled provider was associated with a 90% decrease in the odds of neonatal death.

Our findings on the effect of PNC on neonatal death reflected the general pattern found in an analysis of data from DHS surveys in 10 African countries, which found that PNC by unskilled providers was associated with a 32% decrease in the probability of neonatal death and PNC by a skilled provider with a 60% decrease in the probability of neonatal death (Sing, Brodish, and Haney 2014). Analysis of data from a DHS survey in Nepal did not find a significant association between PNC and neonatal mortality. In this survey, the authors explained that the lack of association could be due to sample size and other limitations in the survey variables (Paudel et al. 2013).

Other research on the relationship between PNC and neonatal death has argued that the influence of PNC on neonatal death may be overstated if early neonatal deaths that occur in the first 24 hours after birth are not excluded (Mallick et al. 2018). Such deaths represented nearly half of all neonatal deaths in our analytic sample. Our sensitivity analysis showed our results to be robust to the inclusion or exclusion of deaths in the first 24 hours. When we exclude deaths in the first 24 hours, the odds of neonatal death decreases by 80% when PNC is provided by a skilled provider, as compared to 90% decreased odds when we do not exclude these deaths. We conclude that we did not overestimate the strength of the association between PNC and neonatal death.

While PNC was low (36% in the first 2 days) in this study, less than half of mothers in the world receive a PNC visit within 2 days of childbirth (Lawn et al. 2014). Lower levels of PNC than in our analysis were found in Sub-Saharan African countries. An analysis of DHS data from 23 Sub-Saharan African countries found that only 13% of women who delivered at home received PNC within 2 days of birth (Warren et al. 2006).

We find that ANC is a significant predictor of PNC by a skilled provider. We did not find such an association with place of delivery. Studies in a range of settings have shown that use of health services earlier in the continuum of care influence PNC use (Aryal et al. 2019; Mbugua and MacQuarrie 2018; Belachew et al. 2016; Owili et al. 2016; Obago 2013). However, these studies typically show that ANC loses its association

22

with PNC after the place of delivery is controlled. This demonstrates that ANC may act on PNC by influencing women to deliver in a health facility. Our study finds that ANC has a strong independent association with PNC regardless of the place of delivery. Our study concurs with a broad literature that indicates the importance of all stages in the continuum of maternal health care.

Our finding that PNC was associated with a low risk of neonatal death suggests that neonate health might have been improved by health services and health messages the caregivers received from providers about newborn care during the postnatal period. ANC, delivery by skilled birth attendants, and PNC are essential elements in an integrated care approach for mothers and neonates.

Providing PNC with a skilled provider is the ideal approach, although PNC with an unskilled provider such as an auxiliary midwife is a useful alternative in very remote areas where PNC by skilled providers is not feasible. Unskilled providers should be given proper, regular training on the key elements of newborn care in order to enhance the beneficial effect of PNC and to know when to refer to a skilled provider. Unskilled providers also need to work closely with the skilled providers. A strong referral system between community and facility care has been recommended as an integrated approach to newborn health (WHO and UNICEF 2009). Training for unskilled providers should emphasize when to refer to a skilled provider.

4.2 Limitations

Our study has some limitations. Since the study is a secondary data analysis, we relied on available variables which are not comprehensive for control of all confounding variables. Thus, some residual confounding may remain. Many women who deliver at home do not know their child’s exact birth weight. We used a rough classification of birth size based on reported birth size as very large, larger than average, average, smaller than average, or very small. Questions on gestational age, another important confounder, are not included in the DHS. Finally, we can only estimate associations, but not causality because this is a cross-sectional survey.

This study also has some strengths. Based on a nationally representative sample, the MDHS is a methodologically rigorous survey, with high validity and reliability of the variables used in the analysis. Thus, we were able to provide evidence that supported efforts to promote PNC as a means of preventing neonatal deaths. This finding is important because countries like Myanmar are considering scaling-up PNC services in order to reach more newborns and save more lives.

4.3 Conclusion

This study is the first to evaluate associations between PNC by provider type and neonatal death in Myanmar. The results show clearly that Myanmar should continue to promote PNC as a means of preventing neonatal mortality and improving child survival. The provision of PNC at the national level is currently low. This study provides evidence for advocating scaling-up of PNC in the country.

23

As Myanmar works to increase the number of midwives as skilled providers of maternal care, provision of PNC by auxiliary midwives with proper training in newborn care may be used as an alternative strategy to improve neonatal death in remote areas where midwives and other skilled providers are not available. ANC and delivery care by skilled birth attendants should also be promoted, along with PNC, as an integrated care approach for both mothers and neonates. This study shows that the major direction for public health interventions to reduce neonatal death is improving the utilization of PNC, which should also be linked with promotion of proper ANC and institutional delivery.

25

REFERENCES

Aryal, K. K., S. K. Sharma, M. N. Khanal, B. Bista, S. L. Sharma, S. Kafle, and M. M. Steffen. 2019. Trends and Determinants of Maternal Health Care in Nepal. DHS Further Analysis Reports No. 118. Rockville, Maryland, USA: ICF.

Belachew, T., A. Taye, and T. Belachew. 2016. “Postnatal Care Services Utilization and Associated Factors among Mothers in Lemo Woreda, Ethiopia.” Journal of Women’s Health Care 5 (3): 1-7. https://www.omicsonline.org/open-access/postnatal-care-service-utilization-and-associated-factors-among-mothersin-lemo-woreda-ethiopia-2167-0420-1,000318.php?aid=74603.

The Republic of the Union of Myanmar, Ministry of Health, Department of Public Health, Child Health Division. 2014. Study on Cause of Under-five Mortality: Ministry of Health and UNICEF. https://www.unicef.org/myanmar/COD_body.pdf.

Hug, L., D. Sharrow, and D. You. 2017. Levels & Trends in Child Mortality: Report 2017. Estimates Developed by the UN Inter-agency Group for Child Mortality Estimation. http://www.everywomaneverychild.org/wp-content/uploads/2017/10/Child_Mortality_Report_2017_ UNICEF-WHO.pdf.

Lawn, Joy E., S. Cousens, J. Zupan, and the Lancet Neonatal Survival Steering Team. 2005. “4 million neonatal deaths: When? Where? Why?” The Lancet 365 (9462): 891-900. https://doi.org/10.1016/S0140-6736(05)71048-5.

Lawn, Joy E., H. Blencowe, S. Oza, D. You, A. C. C. Lee, P. Waiswa, M. Lalli, et al. 2014. “Every Newborn: Progress, Priorities, and Potential Beyond Survival.” The Lancet 384 (9938): https://doi.org/10.1016/S0140-6736(14)60496-7189-205.

Mbugua, S., and K. L. D. MacQuarrie. 2018. Determinants of Maternal Care-Seeking in Kenya. DHS Further Analysis Reports No. 111. Rockville, Maryland, USA: ICF. https://dhsprogram.com/pubs/pdf/FA111/FA111.pdf.

Mallick, Lindsay, J. Yourkavitch, and C. Allen. 2018. Thermal Care and Umbilical Cord Care Practices and their Associations with Newborn Mortality. DHS Analytical Studies 68. Rockville, MD, USA: ICF. https://dhsprogram.com/pubs/pdf/AS68/AS68.pdf.

Ministry of Health and Sports (MoHS) and ICF. 2017. Myanmar Demographic and Health Syrvey 2015-16. Nay Pyi Taw, Myanmar, and Rockville, Maryland USA: Ministry of Health and Sports and ICF. https://dhsprogram.com/pubs/pdf/FR324/FR324.pdf.

Obago, I. T. 2013. The Role of Antenatal Care in Predicting Health Facility Delivery among Women in Kenya: Further Analysis of Data from the 2008-09 KDHS. DHS Working Papers No. 86. Calverton, Maryland, USA: ICF International. https://dhsprogram.com/pubs/pdf/WP86/WP86.pdf.

26

Owili, P. O., M. A. Muga, Y. Chou, Y. E. Hsu, N. Huang, and L-Y. Chien. 2016. “Associations in the Continuum of Care for Maternal, Newborn and Child Health: A Population-Based Study of 12 Sub-Saharan Africa Countries.” BMC Public Health 16 (1):414. https://doi.org/10.1186/s12889-016-3075-0.

Paudel, D., A. Thapa, P. R. Shedain, and B. Paudel. 2013. Trends and Determinants of Neonatal Mortality in Nepal: Further Analysis of the Nepal Demographic and Health Surveys, 2001-2011. Calverton, Maryland, USA: Nepal Ministry of Health and Population, New ERA, and ICF International. https://dhsprogram.com/pubs/pdf/FA75/FA75.pdf.

Sing, K., P. Brodish, and E. Haney. 2014. “Postnatal Care by Provider Type and Neonatal Death in Sub-Saharan Africa: A Multilevel Analysis.” BMC Public Health 14: 941. https://doi.org/10.1186/1471-2458-14-941.

StataCorp. 2017. Stata Statistical Software: Release 15. College Station, TX: StataCorp LLC.

World Health Organization (WHO) and United Nations Children’s Fund (UNICEF). 2009. Home Visits for the Newborn Child: A Strategy to Improve Survival. Geneva: WHO. https://apps.who.int/iris/bitstream/handle/10665/70002/WHO_FCH_CAH_09.02_eng.pdf;jsessionid=015BA9CC03CBAB3D1B4E05F279FEB4C4?sequence=1.

27

APPENDIX

Appendix Table A1 Odds ratio of neonatal death from multiple logistic regression among last children under age 5 born to interviewed women (n=3,426)

Characteristics Odds ratio 95% CI

Residence Urban Ref. Rural 1.2 [0.5,3.1]

Wealth index Poorest 2.1 [0.4,11.4] Poorer 2.8 [0.6,14.2] Middle 1.4 [0.2,8.2] Richer 2 [0.5,8.5] Richest Ref.

Source of water Non-improved 1.1 [0.5,2.4] Improved Ref.

Type of toilet Shared 2.3 [0.7,7.7] Unimproved 1.1 [0.5,2.4] Improved Ref.

Mother’s age at birth of the child <20 years 0.1** [0.0,0.4] 20-29 years Ref. 30-39 years 0.8 [0.4,1.7] 40+ years 0.5 [0.1,2.3]

Mother’s education No education 0.4 [0.1,2.2] Primary 0.2 [0.0,1.1] Secondary 0.3 [0.1,1.4] Higher Ref.

Mother’s smoking status No current smoking Ref. [1.0,1.0] Current smoking 0.8 [0.2,3.3]

Smokers at home of the child Smokers at home 1 [0.5,2.0] No smokers at home Ref.

Sex of the child Male Ref. Female 1.5 [0.8,3.1]

Birth size Very large 3.6 [0.4,35.4] Larger than average 1.7 [0.7,4.0] Average Ref. Smaller than average 4.1** [1.5,11.5] Very small 9.3** [2.1,42.6]

Continued…

28

Appendix Table A1 Continued

Characteristics Odds ratio 95% CI

Previous birth interval1 <2 years 1.1 [0.3,3.7] 2 years 1.4 [0.5,3.8] 3 years 1.4 [0.4,5.4] 4+ years Ref. [1.0,1.0] The first child 0.5 [0.1,1.5]

Birth order 1 Ref. 2-3 1 [1.0,1.0] 4-6 1.8 [0.7,4.3] 7+ 2.2 [0.5,8.9]

Multiple birth Single birth Ref. Multiple birth 4.6 [0.6,33.7]

Number of ANC visits2 <4 visits 0.6 [0.2,1.3] 4+ visits Ref. [1.0,1.0]

Type of delivery Institutional Ref. Non-institutional 1.2 [0.4,3.8]

Postnatal care (PNC) No PNC 8.9*** [2.7,29.8] PNC by non-skilled providers 4.8* [1.1,21.9] PNC by skilled providers Ref.

* p<0.05, ** p<0.01, *** p<0.001 1 “Don’t know” responses were excluded. 2 “Don’t know” and “Missing value” responses were excluded.

Appendix Table A2 Odds ratio of neonatal death by type of PNC provider from bivariate logistic regression

(n=3,561) and multiple logistic regression (n=3,324) among last children under age 5 born to interviewed women, excluding neonatal deaths in the first 24 hours of life

Unadjusted odds ratio 95% CI P value

Adjusted odds ratio1 95% CI P value

Characteristics Bivariate regression Multivariate regression

No PNC 4.7 [1.2,18.2] 0.027 4.8 [1.3,17.5] 0.019 PNC by non-skilled providers 4.4 [0.8,24.43] 0.093 3.4 [0.6,18.3] 0.153 PNC by skilled providers Ref. Ref.

1 Adjusted for place of residence, wealth index, source of water, type of toilet, mother’s age at birth of the child, educational level, current smoking status of mother, smokers in the home, the sex of child, birth size, previous birth interval, birth order, multiple births, number of ANC visits, place of delivery, and type of provider for PNC visit.

29

Appendix Table A3 Odds ratio of PNC from multiple logistic regression, among last children under age 5 born to interviewed women (n=3,444)

Characteristics Odds ratio1 95% CI

Residence Urban Ref. Rural 1.7*** [1.3, 2.3]

Wealth index Poorest Ref. Poorer 1.1 [0.9,1.5] Middle 1.3 [1.0,1.7] Richer 1.2 [0.9,1.7] Richest 1.5 [0.9,2.3]

Source of water Non-improved Ref. Improved 0.9 [0.7,1.1]

Type of toilet Shared 1 [0.8,1.4] Unimproved Ref. Improved 0.8 [0.7,1.0]

Mother’s age at birth of the child <20 years Ref. 20-29 years 1.1 [0.8,1.5] 30-39 years 1.2 [0.8,1.7] 40+ years 1.4 [0.9,2.2]

Mother’s education No education Ref. Primary 1.4* [1.0,1.8] Secondary 1.3 [1.0,1.8] Higher 1.6* [1.1,2.4]

Mother’s smoking status Current smoking Ref. No current smoking 1.3 [0.9,1.9]

Smokers at home of the child Smokers at home Ref. No smokers at home 1.1 [0.9,1.3]

Sex of the child Male Ref. Female 1.1 [0.9,1.2]

Birth size Very large 1.4 [0.8,2.4] Larger than average 1.2 [1.0,4.1.5] Average Ref. Smaller than average 1 [0.8,1.2] Very small 1 [0.5,1.8]

Continued…

30

Appendix Table A3 Continued

Characteristics Odds ratio1 95% CI

Previous birth interval2 <2 years Ref. 2 years 1.3 [0.9,1.9] 3 years 1.2 [0.8,1.7] 4+ years 1.2 [0.8,1.6] The first child 1.4* [1.0,2.0]

Birth order 2-3 Ref. 4-6 1.1 [0.8,1.3] 7+ 0.9 [0.6,1.4]

Multiple birth Single birth Ref. Multiple birth 2.1 [0.7,6.1]

Number of ANC visits3 <4 visits Ref. 4+ visits 1.7*** [1.4,2.4]

Type of delivery Non-institutional Ref. Institutional 0.7 [0.6,0.9]

* p<0.05, ** p<0.01, *** p<0.001 1 Adjusted for place of residence, wealth index, source of water, type of toilet, mother’s age at birth of the child, educational level, current smoking status of mother, smokers in the home, the sex of child, birth size, previous birth interval, birth order, multiple births, number of ANC visits, place of delivery, type of provider for PNC visit 2 “Don’t know” responses were excluded. 3 “Don’t know” and “Missing value” responses were excluded.